--cpu=name


This option enables code generation for the selected ARM processor or architecture.

Syntax

--cpu=name

Where:

name

is the name of a processor or architecture.

If name is the name of a processor, enter it as shown on ARM data sheets, for example, ARM7TDMI, ARM1176JZ-S, MPCore.

If name is the name of an architecture, it must belong to the list of architectures shown in Table 3.

Processor and architecture names are not case-sensitive.

Wildcard characters are not accepted.

Supported ARM architectures

Architecture Name	Description	Example processors
4	ARMv4 without Thumb	SA-1100
4T	ARMv4 with Thumb	ARM7TDMI, ARM9TDMI, ARM720T, ARM740T, ARM920T, ARM922T, ARM940T, SC100
5T	ARMv5 with Thumb and interworking
5TE	ARMv5 with Thumb, interworking, DSP multiply, and double-word instructions	ARM9E, ARM946E-S, ARM966E-S
5TEJ	ARMv5 with Thumb, interworking, DSP multiply, double-word instructions, and Jazelle®extensions[a]	ARM926EJ-S, ARM1026EJ-S, SC200
6	ARMv6 with Thumb, interworking, DSP multiply, double-word instructions, unaligned and mixed-endian support, Jazelle, and media extensions	ARM1136J-S, ARM1136JF-S
6-M	ARMv6 micro-controller profile with Thumb only plus processor state instructions	Cortex-M1 without OS extensions, Cortex-M0, SC000
6S-M	ARMv6 micro-controller profile with Thumb only, plus processor state instructions and OS extensions	Cortex-M1 with OS extensions
6K	ARMv6 with SMP extensions	MPCore
6T2	ARMv6 with Thumb-2	ARM1156T2-S, ARM1156T2F-S
6Z	ARMv6 with Security Extensions	ARM1176JZF-S, ARM1176JZ-S
7	ARMv7 with Thumb-2 only and without hardware divide	Cortex-A5
7-A	ARMv7 application profile supporting virtual MMU-based memory systems, with ARM, Thumb-2, and Thumb-2EE instruction sets, DSP support, and 32-bit SIMD support	Cortex-A8, Cortex-A9
7-A.security	Enables the use of the SMCinstruction (formerly SMI) when assembling for the v7-A architecture	Cortex-A5, Cortex-A8, Cortex-A9
7-R	ARMv7 real-time profile with ARM, Thumb-2, DSP support, and 32-bit SIMD support	Cortex-R4, Cortex-R4F
7-M	ARMv7 micro-controller profile with Thumb-2 only and hardware divide	Cortex-M3, SC300
7E-M	ARMv7-M enhanced with DSP (saturating and 32-bit SIMD) instructions	Cortex-M4
[a] The ARM compiler cannot generate Java bytecodes.

Note

ARMv7 is not an actual ARM architecture.

--cpu=7 denotes the features that are common to all of the ARMv7-A, ARMv7-R, and ARMv7-M architectures.

By definition, any given feature used with --cpu=7 exists on all of the ARMv7-A, ARMv7-R, and ARMv7-M architectures.

7-A.security is not an actual ARM architecture, but rather, refers to 7-Aplus Security Extensions.

Default

If you do not specify a --cpu option, the compiler assumes --cpu=ARM7TDMI.

To obtain a full list of CPU architectures and processors, use the --cpu=listoption.

Usage

The following general points apply to processor and architecture options:

Processors

• Selecting the processor selects the appropriate architecture,Floating-Point Unit (FPU), and memory organization.
• The supported --cpu values include all current ARM product names or architecture versions.

  Other ARM architecture-based processors, such as the Marvell Feroceon and the Marvell XScale, are also supported.

• If you specify a processor for the --cpu option, the compiled code is optimized for that processor.
  This enables the compiler to use specific coprocessors or instruction scheduling for optimum performance.

Architectures

• If you specify an architecture name for the --cpu option, the code is compiled to run on any processor supporting that architecture.
  For example, --cpu=5TE produces code that can be used by theARM926EJ-S^®.

FPU

• Some specifications of --cpu imply an --fpu selection.
  For example, when compiling with the --arm option, --cpu=ARM1136JF-S implies --fpu=vfpv2.
  Similarly, --cpu=Cortex-R4F implies --fpu=vfpv3_d16.

  Note

  Any explicit FPU, set with --fpu on the command line, overrides an implicit FPU.

• If no --fpu option is specified and no --cpu option is specified, --fpu=softvfp is used.

ARM/Thumb

• Specifying a processor or architecture that supports Thumb instructions,
  such as --cpu=ARM7TDMI, does not make the compiler generate Thumb code.
  It only enables features of the processor to be used, such as long multiply.
  Use the --thumb option to generate Thumb code, unless the processor is a Thumb-only processor, for example Cortex-M4.
  In this case, --thumb is not required.

  Note

  Specifying the target processor or architecture might make the object code generated
  by the compiler incompatible with other ARM processors.
  For example, code compiled for architecture ARMv6 might not run on an ARM920T processor,
  if the compiled code includes instructions specific to ARMv6.
  Therefore, you must choose the lowest common denominator processor suited to your purpose.

• If you are compiling code that is intended for mixed ARM/Thumb systems for processors that support ARMv4T or ARMv5T,
  then you must specify the interworking option --apcs=/interwork.
  By default, this is enabled for processors that support ARMv5T or above.
• If you compile for Thumb, that is with the --thumb option on the command line,
  the compiler compiles as much of the code as possible using the Thumb instruction set.
  However, the compiler might generate ARM code for some parts of the compilation.
  For example, if you are compiling code for a Thumb-1 processor and using VFP,
  any function containing floating-point operations is compiled for ARM.
• If the architecture you are compiling code for only supports Thumb, there is no need to specify --thumb on the command line.
  For example, if compiling code for ARMv7-M with --cpu=7-M, you do not have to specify --thumb on the command line,
  because ARMv7-M only supports Thumb-2.
  Similarly, ARMv6-M and other Thumb-only architectures.

Restrictions

You cannot specify both a processor and an architecture on the same command-line.

--fpu=name

This option enables you to specify the target FPU architecture.

If you specify this option, it overrides any implicit FPU option that appears on the command line, for example, where you use the --cpu option.

To obtain a full list of FPU architectures use the --fpu=list option.

Syntax

--fpu=name

Where name is one of:

none

Selects no floating-point option. No floating-point code is to be used. This produces an error if your code contains float types.

vfpv

This is a synonym for vfpv2.

vfpv2

Selects a hardware vector floating-point unit conforming to architecture VFPv2.

Note

If you enter armcc --thumb --fpu=vfpv2 on the command line,
the compiler compiles as much of the code using the Thumb instruction set as possible,
but hard floating-point sensitive functions are compiled to ARM code.
In this case, the value of the predefine __thumb is not correct.

vfpv3

Selects a hardware vector floating-point unit conforming to architecture VFPv3.
VFPv3 is backwards compatible with VFPv2 except that VFPv3 cannot trap floating-point exceptions.

vfpv3_fp16

Selects a hardware vector floating-point unit conforming to architecture VFPv3
that also provides the half-precision extensions.

vfpv3_d16

Selects a hardware vector floating-point unit conforming to VFPv3-D16 architecture.

vfpv3_d16_fp16

Selects a hardware vector floating-point unit conforming to VFPv3-D16 architecture,
that also provides the half-precision extensions.

vfpv4

Selects a hardware floating-point unit conforming to FPv4 architecture.

vfpv4_d16

Selects a hardware floating-point unit conforming to the VFPv4-D16 architecture.

fpv4-sp

Selects a hardware floating-point unit conforming to the single precision variant of the FPv4 architecture.

softvfp

Selects software floating-point support where floating-point operations are performed by a floating-point library, fplib.
This is the default if you do not specify a --fpu option, or if you select a CPU that does not have an FPU.

softvfp+vfpv2

Selects a hardware vector floating-point unit conforming to VFPv2, with software floating-point linkage.
Select this option if you are interworking Thumb code with ARM code on a system that implements a VFP unit.

If you select this option:

• Compiling with --thumb behaves in a similar way to --fpu=softvfp
except that it links with floating-point libraries that use VFP instructions.
• Compiling with --arm behaves in a similar way to --fpu=vfpv2
except that all functions are given software floating-point linkage.
  This means that functions pass and return floating-point arguments and
  results in the same way as --fpu=softvfp, but use VFP instructions internally.

Note

If you specify softvfp+vfpv2 with the --arm or --thumb option for C code,
it ensures that your interworking floating-point code is compiled to use software floating-point linkage.

softvfp+vfpv3

Selects a hardware vector floating-point unit conforming to VFPv3, with software floating-point linkage.
Select this option if you are interworking Thumb code with ARM code on a system that implements a VFPv3 unit.

softvfp+vfpv3_fp16

Selects a hardware vector floating-point unit conforming to VFPv3-fp16, with software floating-point linkage.

softvfp+vfpv3_d16

Selects a hardware vector floating-point unit conforming to VFPv3-D16, with software floating-point linkage.

softvfp+vfpv3_d16_fp16

Selects a hardware vector floating-point unit conforming tovfpv3_d16_fp16, with software floating-point linkage.

softvfp+vfpv4

Selects a hardware floating-point unit conforming to FPv4, with software floating-point linkage.

softvfp+vfpv4_d16

Selects a hardware floating-point unit conforming to VFPv4-D16, with software floating-point linkage.

softvfp+fpv4-sp

Selects a hardware floating-point unit conforming to FPv4-SP, with software floating-point linkage.

Usage

Any FPU explicitly selected using the --fpu option always overrides any FPU implicitly selected using the --cpu option.
For example, the option --cpu=ARM1136JF-S --fpu=softvfp generates code that uses the software floating-point library fplib,
even though the choice of CPU implies the use of architecture VFPv2.

To control floating-point linkage without affecting the choice of FPU, you can use --apcs=/softfp or --apcs=/hardfp.

Restrictions

The compiler only permits hardware VFP architectures (for example, --fpu=vfpv3, --fpu=softvfp+vfpv2),
to be specified when MRRC and MCRR instructions are supported in the processor instruction set. 
MRRC and MCRR instructions are not supported in 4, 4T, 5T and 6-M.
Therefore, the compiler does not allow the use of these CPU architectures with hardware VFP architectures.

Other than this, the compiler does not check that --cpu and --fpu combinations are valid.
Beyond the scope of the compiler, additional architectural constraints apply.
For example, VFPv3 is not supported with architectures prior to ARMv7.
Therefore, the combination of --fpu and --cpu options permitted by the compiler does not necessarily translate to the actual device in use.

The compiler only generates scalar floating-point operations.
If you want to use VFP vector operations, you must do this using assembly code.

NEON support is disabled for softvfp.

Default

The default target FPU architecture is derived from use of the --cpu option.

If the CPU specified with --cpu has a VFP coprocessor, the default target FPU architecture is the VFP architecture for that CPU.
For example, the option --cpu ARM1136JF-S implies the option --fpu vfpv2.
If a VFP coprocessor is present, VFP instructions are generated.

If you are building ARM Linux applications with --arm_linux or --arm_linux_paths,
the default is always software floating-point linkage.

Even if you specify a CPU that implies an FPU (for example, --cpu=ARM1136JF-S),
the compiler still defaults to --fpu=softvfp+vfp, not --fpu=vfp.

If there is no VFP coprocessor, the compiler generates code
that makes calls to the software floating-point library fplib to carry out floating-point operations.